Search Images Maps Play YouTube News Gmail Drive More »
Sign in
Screen reader users: click this link for accessible mode. Accessible mode has the same essential features but works better with your reader.

Patents

  1. Advanced Patent Search
Publication numberUS4161101 A
Publication typeGrant
Application numberUS 05/722,066
Publication dateJul 17, 1979
Filing dateSep 9, 1976
Priority dateSep 9, 1976
Publication number05722066, 722066, US 4161101 A, US 4161101A, US-A-4161101, US4161101 A, US4161101A
InventorsThane M. Drummond
Original AssigneeGeneral Electric Company
Export CitationBiBTeX, EndNote, RefMan
External Links: USPTO, USPTO Assignment, Espacenet
Control system for and method of zero error automatic calibration of gas turbine temperature control parameters
US 4161101 A
Abstract
A control system including a controller for reading temperature values from sensors in a gas turbine, determining the reliability of the sensors, and issuing a fuel control signal for use in controlling the turbine is disclosed. The fuel control signal has a calculated value which is a function of the turbine operating temperature derived from the values of temperature signals generated by the sensors.
Images(21)
Previous page
Next page
Claims(8)
What is claimed is: ##SPC1## ##SPC2## ##SPC3## ##SPC4##
1. In a control system of the type including a controller for providing a fuel control signal to a gas turbine for controlling fuel flow thereto to control the temperature of said turbine, said fuel control signal having a value derived as a function of the average of the values of temperature signals provided to the controller from sensors in the turbine, the improvement in said control system comprising:
(a) means for specifying to said controller a reference having a value proportional to a desired turbine operating temperature;
(b) means in said controller for deriving a difference value proportional to the difference in magnitude between said reference and the value derived from the average of the temperature signals;
(c) means in said controller for developing a first adjustment value which varies at a rate specified by a vernier constant when said difference value exceeds a predetermined magnitude;
(d) means in said controller for developing a second adjustment value, said second adjustment value having a magnitude proportional to the difference in the value of a mid-value signal and the value derived from the average values of the temperature signals;
(e) means in said controller responsive to said first adjustment value and to said second adjustment value for developing a temperature trim signal proportional to the difference therebetween; and,
(f) means for providing said fuel control signal to said turbine including,
(1) means responsive to specified ones of the temperature signals for generating said mid-value signal, said mid-value signal having a magnitude intermediate the individual values of the temperature signals provided by said specified ones, and
(2) means for combining said mid-value signal and said temperature trim signal to thereby provide said fuel control signal.
2. In a control system of the type including a controller for providing a fuel control signal to a gas turbine for controlling fuel flow thereto to control the temperature of said turbine, said fuel control signal having a value derived as a function of the average of the values of temperature signals provided to the controller from sensors in the turbine, the improvement in said control system comprising:
(a) means for specifying to said controller a first reference having a value proportional to a desired turbine operating temperature;
(b) means in said controller for generating a difference signal having a value proportional to the difference between said first reference and an average value derived from the average of the temperature signals;
(c) means in said controller responsive to said difference signal for generating a first correction signal which varies at a rate specified by a vernier constant when a difference exists between said first reference and said average value;
(d) means for specifying to said controller a second reference having a value proportional to a safe turbine operating temperature;
(e) means in said controller responsive to said first and second references for generating a temperature bias reference having a value proportional to the difference between said first and second references;
(f) means for combining said average value and a mid-value signal to generate a second correction signal;
(g) means for algebraically combining said first correction signal, said temperature bias reference and said second correction signal to generate a temperature trim signal; and,
(h) means including,
(1) means responsive to specified ones of the temperature signals for generating said mid-value signal, said mid-value signal having a magnitude intermediate the individual values of the temperature signals provided by said specified ones, and,
(2) means for algebraically combining said mid-value signal, said temperature trim signal and said second reference for generating the fuel control signal provided to said turbine.
3. An improved method of controlling the operating temperature of a gas turbine in a system of the type including a controller for providing a fuel control signal to said turbine for controlling fuel flow thereto, said fuel control signal having a value derived as a function of the average of the values of temperature signals provided to the controller from sensors in the turbine, the method comprising the steps of:
(a) specifying to said controller first and second references, said first reference having a value proportional to a desired turbine operating temperature and said second reference having a value proportional to a safe turbine operating temperature;
(b) developing in said controller a first correction value which varies in magnitude at a rate specified by a vernier constant when a difference exists between said first reference and an average value derived from the average values of the temperature signals;
(c) algebraically combining in said controller said first and second references to generate a temperature bias reference;
(d) providing to said controller a mid-value signal having a value mid the individual values of specified ones of the temperature signals;
(e) algebraically combining in said controller said average value and said mid-value signal to generate a second correction value;
(f) algebraically combining in said controller said first and second correction values and said temperature bias reference to generate a temperature trim signal by said controller; and,
(g) algebraically combining said mid-value signal, said temperature trim signal and said second reference to provide said fuel control signal to said turbine.
4. In a control system, a method of controlling the operating temperature of a gas turbine in accordance with turbine temperature represented by the values of temperature signals, at least some of said temperature signals, provided to a computer from temperature sensors in the turbine, comprising the steps of:
(a) providing to the computer first and second reference signals and a mid-value signal, said first reference signal having a value proportional to a desired turbine operating temperature, said second reference signal having a value proportional to a safe turbine operating temperature, and said mid-value signal having a value intermediate the individual values of specified ones of the temperature signals;
(b) storing in the computer a program of instructions, said program directing the computer to,
(1) calculate an average value proportional to the average turbine temperature in accordance with the values of the temperature signals provided to the computer,
(2) calculate the difference between said first reference and said average value,
(3) develop a first correction value having a rate of change specified by a vernier constant when a difference exists between said first reference and said average value,
(4) develop a temperature bias reference having a value proportional to the difference between said first and second references,
(5) develop a second correction value proportional to the difference between said average value and said mid-value signal,
(6) calculate the value of a temperature trim signal for output by said computer, said temperature trim signal having a value proportional to the sum of said first and second correction values and said temperature bias reference; and,
(c) algebraically summing said temperature trim signal, said mid-value signal and said second reference to provide a fuel control signal to said gas turbine to control the flow of fuel to the turbine.
5. In a control system, a method of controlling the operating temperature of a gas turbine in accordance with turbine temperature represented by the values of temperature signals provided to a computer from temperature sensors in the turbine, comprising the steps of:
(a) providing to the computer a reference signal having a value proportional to a desired turbine operating temperature;
(b) providing to the computer a mid-value signal having a value intermediate the individual values of specified ones of the temperature signals;
(c) providing to the computer at least some of said temperature signals;
(d) storing in the computer an executable program of instructions, said program, when executed, directing said computer to,
(1) calculate a value proportional to the average turbine temperature in accordance with the average of the values of the temperature signals provided to the computer,
(2) develop a correction value proportional to the difference in magnitude between the value of said reference signal and the value proportional to the average turbine temperature,
(3) adjust the correction value by a predetermined vernier constant for each execution of said program if a difference exists between the value of said reference signal and the value proportional to the average turbine temperature,
(4) calculate the value of a temperature trim signal for output by said computer, said temperature trim signal having a value derived from the sum of said correction value and the difference in magnitude between said mid-valve signal and the value proportional to the average turbine temperature; and,
(e) algebraically combining said temperature trim signal and said mid-value signal to provide a fuel control signal to said gas turbine to control the flow of fuel to the turbine and the operating temperature thereof.
6. A system for controlling fuel flow to a gas turbine in accordance with turbine temperature represented by the values of temperature readings provided by sensors in the turbine, said system comprising:
(a) means for providing a predetermined reference proportional to a desired turbine operating temperature; and sensor means for providing the values of said temperature readings
(b) a programmable controller including,
(1) a storage means for retaining a program of instructions and holding data related to the calculation of a temperature trim signal for output by said controller,
(i) certain ones of said instructions directing said controller to determine if the value of each temperature reading is within prescribed limits and to reject those temperature readings which are out of limits,
(ii) certain other ones of said instructions directing said controller to calculate the value of said temperature trim signal, said temperature trim signal having a value proportional to the sum of first and second adjustment values, said first adjustment value varying at a rate specified by a vernier constant when there is a difference in the magnitude between said reference and a calculated value proportional to the average of unrejected temperature readings and said second adjustment value having a magnitude proportional to the difference in the value of a mid-value signal applied to said controller and said calculated value; and,
(c) means for providing a fuel control signal to said turbine for controlling fuel flow thereto including,
(1) means coupled to specified ones of said sensors and responsive to the temperature readings provided thereby for generating said mid-value signal, said mid-value signal having a magnitude substantially proportional to the average turbine operating temperature, and
(2) means for combining said mid-value signal and said temperature trim signal to thereby provide said fuel control signal.
7. In a control system of the type including a controller for providing a fuel control signal to a gas turbine for controlling fuel flow thereto to control the temperature of said turbine, said fuel control signal having a value derived as a function of the average of the values of temperature signals provided to the controller from sensors in the turbine, the improvement in said control system comprising:
(a) means in said controller including,
(1) select means responsive to said temperature signals to determine the reliability of each sensor and to allow at the output thereof only those temperature signals which are within prescribed limits, and
(2) averaging means responsive to the temperature signals at the output of said select means for generating a temperature average signal proportional to the average of the temperature signals applied thereto;
(b) means for specifying to said controller a reference signal having a value proportional to a desired turbine operating temperature;
(c) means in said controller responsive to said reference signal and said temperature average signal for generating a first correction signal proportional to the difference in magnitude between said reference and temperature average signals;
(d) means in said controller responsive to said first correction signal for developing an adjustment signal which varies at a specified rate when said first correction signal exceeds a predetermined magnitude;
(e) means in said controller responsive to a mid-value signal applied to said controller and said temperature average signal for generating a second correction signal proportional to the difference in the magnitude therebetween;
(f) means in said controller responsive to said adjustment signal and said second correction signal for developing a temperature trim signal proportional to the sum thereof; and,
(g) means for providing said fuel control signal to said turbine including,
(1) means connected to specified ones of said sensors and responsive to the temperature signals provided thereby for generating said mid-value signal, said mid-value signal having a magnitude substantially proportional to the average turbine operating temperature, and,
(2) means for combining said mid-value signal and said temperature trim signal to thereby provide said fuel control signal.
8. In a control system of the type including a controller for providing a fuel control signal to a gas turbine for controlling fuel flow thereto to control the temperature of said turbine, said fuel control signal having a value derived as a function of the average of the values of temperature signals provided to the controller from sensors in the turbine, the improvement in said control system comprising:
(a) means for specifying to said controller a first reference signal having a value proportional to a desired turbine operating temperature;
(b) means for specifying to said controller a second reference signal having a value proportional to a safe turbine operating temperature;
(c) means in said controller including,
(1) select means resonsive to said temperature signals to determine the reliability of each sensor to allow at the output thereof only those temperature signals which are within prescribed limits,
(2) averaging means responsive to the temperature signals at the output of said select means for generating a temperature average signal proportional to the average of the temperature signals applied thereto;
(d) means in said controller responsive to said first refernce signal and said temperature average signal for generating a first correction signal proportional to the difference in magnitude between said first reference and temperature average signals;
(e) means in said controller responsive to said first correction signal for developing an adjustment signal which varies at a specified rate when said first correction signal exceeds a predetermined magnitude;
(f) means in said controller responsive to a mid-value signal applied to said controller and said temperature average signal for generating a second correction signal proportional to the difference in the magnitude therebetween;
(g) means in said controller responsive to said first and second reference signals for generating a third reference signal proportional to the difference in the magnitude between said first and second reference signals;
(h) means in said controller responsive to said adjustment signal, said second correction signal and said third reference signal for developing a temperature trim signal proportional to the sum thereof; and,
(i) means for providing said fuel control signal to said turbine including,
(1) means connected to specified ones of said sensors and responsive to the temperature signals provided thereby for generating said mid-value signal, said mid-value signal having a magnitude substantially proportional to the average turbine operating temperature, and,
(2) means for combining said mid-value signal, said temperature trim signal and said second reference signal to thereby provide said fuel control signal.
Description
BACKGROUND OF THE INVENTION

1. Field of the Invention

This invention relates generally to gas turbine control systems and more particularly to an improved control system and method of controlling fuel flow to a gas turbine to control the temperature thereof.

2. Description of the Prior Art

The present day high acceptance of large capacity gas turbine power plants for large peak load applications has made the reliability and economical operation of these power plants an important concern to the user. This is due in large part to the fact that these power plants are often located in remote areas and operate unattended and to assure higher turbine reliability and economical operation, turbine temperature regulation must be accurate. Remote control is sometimes provided by a telemetry communication link.

Traditionally, the operating temperature of a gas turbine has been controlled by regulating either the firing or exhaust temperature of the turbine. Regardless of which is to be regulated, multiple thermocouples or other temperature sensing devices are employed for sensing the turbine temperature. In most all turbines, to measure the exhaust temperature, a plurality of thermocouples are placed at various points in the turbine exhaust plenum and the temperature signals generated by the thermocouples are averaged to calculate a value proportional to the average exhaust temperature.

In the prior art, this averaging has been done by electronic analog elements interconnected as a temperature regulating system to control the average exhaust temperature at some predetermined level.

The analog elements in temperature regulators, whether they be pure analog or a combination of digital and analog, present several problems:

1. A temperature reference set point or base temperature, usually an analog reference voltage, is employed in controllers for comparison with the average exhaust temperature to derive a fuel control signal for controlling fuel flow to the turbine and hence the operating temperature. It is well known in the art that these analog reference voltages are susceptible to drift with time and/or temperature, thus leading to inaccuracies in temperature control.

2. In digital controllers tolerance and drift errors are introduced into the temperature calculations by the analog-to-digital and digital-to-analog conversion circuitry. Also, round off and truncation in digital calculations can introduce errors or inaccuracies in the temperature calculation.

3. Additional inaccuracies are also introduced in the temperature calculation due to summing resistor tolerances at the inputs to amplifiers. Inaccuracies are also introduced by various buffer amplifiers which are used for signal conditioning of input and output signals of the system controller.

The above inaccuracies all contribute to a disagreement between the aforementioned temperature reference set point and the calculated average turbine exhaust temperature. This disagreement accordingly, leads to inaccurate turbine temperature control.

A disagreement between the temperature reference set point and the calculated or determined average turbine temperature can be detrimental to overall system operation. For example, the loss of accuracy of only a few degrees in turbine temperature can degrade both turbine life and fuel economy. Each of these can contribute to a lack of customer or user acceptance of a system.

It is, therefore, desirable to provide a control system and method for accurately controlling the operating temperature of a gas turbine which enhances turbine life and reduces the operating fuel costs of the turbine.

SUMMARY OF THE INVENTION

The foregoing disadvantages are overcome by the present invention by the provision of a turbine temperature regulating system capable of automatically calibrating out electrical tolerance and drift errors in a fuel control signal for controlling fuel flow to the turbine and hence the temperature thereof.

Control of gas turbine temperature, in one embodiment of the present invention, is afforded by a stored program digital computer or programmable controller which effects the reading and storing of temperature readings or values provided by signals from temperature sensors in the turbine. The computer program is comprised of instructions which, when executed by the computer, calculate the value of a trim signal for use in controlling fuel flow to the turbine and, hence, the turbine operating temperature.

In one instance, the trim signal takes on a value derived from the difference in the magnitude of a specified temperature reference proportional to a desired turbine operating temperature and a calculated value which is a function of turbine operating temperature derived from an averaging of the values of the temperature sensor readings. The trim signal is adjusted at a slow rate by a vernier constant when a difference exists between the temperature reference and the average turbine temperature to close the difference therebetween to calibrate out any electrical tolerance or drift errors in the system causing that difference. By keeping the difference between the temperature reference and the average temperature at zero or within a narrow deadband region, turbine temperature control accuracy is enhanced.

An analog backup means is also implemented into the system to operate the turbine at reduced temperature as specified by a safe turbine temperature reference in the event of computer failure, potential turbine combustion problems, or to trip the turbine, shutting it down, if a serious turbine combustion or instrumentation problem is detected.

Temperature signals from specified ones of the turbine sensors are provided to the analog backup means. The analog means generates a mid-value signal having a magnitude intermediate the values of the specified temperature signals which is approximately equal to the average turbine exhaust temperature. This mid-value signal is used by the computer in calculating the trim signal. Further, the mid-value signal is summed with the safe temperature reference and the trim signal to develop a temperature fuel control signal from the analog means for use in controlling fuel flow to the turbine.

In a second embodiment of the invention, an analog controller continuously monitors the signals or temperature readings generated by the turbine sensors. Means in the analog controller averages the temperature readings to derive a value proportional to the actual turbine temperature. The actual turbine temperature value is algebraically combined with a reference proportional to a desired turbine temperature to derive a correction signal which is slowly adjusted by a vernier constant. The correction signal is combined with a signal proportional to the desired turbine temperature to generate the aforementioned trim signal for application to the analog backup means.

In a third embodiment of the invention, a static digital controller continuously scans the temperature sensors, performs an analog to digital conversion on each sensor temperature reading and calculates the value of the temperature fuel control signal for use in controlling the turbine.

In this latter embodiment, means in the controller enables the calculation of the average turbine temperature from the temperature sensor readings. A storage means receives a value proportional to each temperature reading.

The values of the temperature readings are summed and the value of the actual average turbine temperature is calculated by dividing the number of temperature readings or sensors into the sum of those readings.

The development of a correction signal for calibrating out the previously mentioned errors is calculated by subtracting the actual average temperature from a base reference proportional to a desired turbine operating temperature to derive a difference signal. If a difference exists between the difference signal and a specified reference, the correction signal is adjusted by a vernier constant. This correction signal is combined with the aforementioned base reference to develop the trim signal.

The previously mentioned analog backup means is also implemented in this embodiment. The temperature fuel control signal to the turbine is generated by the analog backup means by combining the trim signal, the mid-value signal and the safe analog temperature reference as previously described.

In each of the above described embodiments, means in the controller detects an out of limits condition for the correction signal and sounds an alarm signifying that condition.

In view of the foregoing, it is therefore an object of the present invention to provide a turbine temperature control system and method having enhanced operating capabilities.

It is another object to provide a stored program computer or controller for calculating the temperature of a gas turbine to develop a correction signal for automatically calibrating out system errors which can introduce inaccuracies in the turbine temperature.

A still further object is to provide a control system and method of providing a fuel control signal to a gas turbine for controlling the temperature thereof whereby the fuel control signal is adjusted by a vernier constant to calibrate out system errors when there is a difference between the average turbine temperature and a specified reference proportional to a desired turbine operating temperature.

It is another object to provide a control system including either a digital or an analog controller for providing a fuel control signal to a gas turbine for controlling the temperature thereof whereby the fuel control signal is adjusted by a vernier constant when there is a difference between the average turbine temperature and a specified reference proportional to a desired turbine operating temperature to thereby remove the difference between the turbine temperature and the reference.

The foregoing and other objects will become apparent as the description proceeds and the features of novelty which characterize the invention will be pointed out in particularity in the claims annexed to and forming a part of this specification.

BRIEF DESCRIPTION OF THE DRAWING

The present invention may be more readily understood by reference to the accompanying drawing in which:

Fig. 1 is a major block diagram of a gas turbine temperature control system in accordance with the present invention.

FIG. 2 is a schematic block diagram of one embodiment of the analog backup means of FIG. 1.

FIG. 3 is a memory map showing in exemplary form the organization of programs and data in the memory of the programmable controller of FIG. 1.

FIGS. 4 through 8 are functional flow charts of an exemplary program for carrying out one embodiment of the present invention.

FIG. 9 is a legendary drawing showing the interrelationships of FIGS. 9a and 9b.

FIGS. 9a and 9b collectively comprise a block diagram schematic of an analog controller as another embodiment of the invention.

FIG. 10 is a legendary drawing showing the interrelationships between FIGS. 10a through 10e.

FIGS. 10a through 10e collectively comprise a block diagram schematic of a digital controller as a further embodiment of the invention.

FIG. 11 is a legendary drawing showing the interrelationships between FIGS. 11a through 11d.

FIGS. 11a through 11d collectively comprise a timing diagram useful in understanding the operation of the controller of FIGS. 10a through 10d.

FIG. 12 is a block diagram schematic illustrating the implementation of a different embodiment of the analog backup means into the digital controller of FIGS. 10a through 10e.

DESCRIPTION OF THE PREFERRED EMBODIMENT

Referring to FIG. 1 of the drawing, a simplified representation of a single shaft gas turbine shown generally as 10 includes a compressor 12, combustion chamber 14, and turbine 16 connected to drive a load 18. Air entering a compressor inlet at 20 supports the combustion of fuel injected by a nozzle 22. The heated exhaust gasses exit from a turbine outlet 24 past distributed temperature sensors or thermocouples 26, the latter serving to generate signals for measuring exhaust temperature in an exhaust plenum 28. A shaft driven variable delivery fuel pump 39 delivers fuel to nozzle 22 at a rate which depends both upon the rotational shaft speed of the gas turbine 10 and the pump stroke setting specified by a fuel control servo 32 mechanically linked to the pump 30. The details of the fuel control servo 32 are not material to the present invention since it may include any servo mechanism which positions the stroke setting control on the pump at a position corresponding to an electrical positioning or fuel control signal (VCE), there being many devices on the market suitable for such use. Also, since the gas turbine may utilize or gas fuels, a gas valve positioner could be employed to control fuel flow to the combustion chamber. The term fuel control signal is used generically herein to control fuel flow to the turbine regardless of the type of fuel used.

The fuel control signal VCE is a single valued electrical signal applied to the fuel control servo 32 by a low value gate 34. The VCE signal always has a value obtained from one of a number of input signals applied to gate 34, whereby each is intended to have a preference over a certain phase of gas turbine operation. The structure and principles of operation of the low value gate 34 ae documented in detail in U.S. Pat. No. 3,520,133, entitled "Gas Turbine Control System" to A. Loft et al, and assigned to the assignee of the present invention.

There are shown in FIG. 1, three input signals provided to the low value gate 34. One of these signals is a temperature fuel control signal TFC provided from a temperature analog backup means 36 forming a part of the invention. Two other signals are applied from a speed/load control 38 and a start-up control 40. In the basic operation of the low value gate 34, its output signal VCE will always represent the value of the lowest input signal applied to that gate. That is, the fuel control signal (VCE) can be no higher than the lowest control signal supplied to gate 34. In the present embodiment the speed/load control 38 and the start-up control 40 form no part of the invention. They are merely included in FIG. 1 to more appropriately put the temperature analog backup means 36 into an overall turbine control environment for clarity and completeness of disclosure. The speed/load control 38 and the start-up 40 may be free standing units receiving input signals from various sensors in the turbine or they may receive calculated signals from a programmable controller or computer 42 via conductors 44 and 46 if the system is under computer control.

The controller 42 may be any one of several well-known types of process computers such as the General Electric 4010 or 4020 process computers, or a programmable controller such as that manufactured by General Electric Company as "Directo-Matic" logic controller. This latter controller is ideally suited for implementation into the preferred embodiment of the present invention and the details of operation of that controller are disclosed in referenced U.S. Pat. No. 3,969,722, issued July 13, 1976, entitled "Method and Apparatus for Operator Interrogation of Simulated Control Circuits" to William J. Danco et al and assigned to the assignee of the present invention.

The controller 42 is preferably of the industrial type and includes analog to digital conversion means for converting applied analog input signals to digital values for storage in the controller memory and digital to analog conversion means for converting digital values to analog signals for output from the controller memory. Since the techniques of converting analog and digital data are well known in the art, the circuitry for performing these functions is not shown. Analog input signals ATCX1-ATCX12 and AOTA-AOTC are provided as temperature readings to the digital controller via a plurality of conductors 48 from the turbine sensors 26. These signals possess values representative of the temperature of the exhaust plenum 28. Further, should a temperature sensor 26 fail, such as by an open or short, the value of the signal on the line corresponding to the failed sensor will provide an invalid condition to the controller indicating that the sensor has failed.

It will be noted that a plurality of specified sensor signals, as shown by the three signals AOTA, AOTB and AOTC, are provided as inputs to the temperature analog backup means 36 via conductors 50. Although these three temperature signals figure into the overall averaging of the turbine temperature, each signals sensor is also utilized as a backup for increasing the realiability of the temperature control system of the present invention. The purpose of these latter temperature signals will become more apparent as the description proceeds in connection with the temperature analog backup means 36.

The analog backup means 36 also receives a temperature trim signal or control signal BRMTEM from the programmable digital controller 42 on a conductor 52. The previously mentioned mid-value signal generated by the analog backup means is shown applied to controller 42 as a signal ATXD on conductor 54. The value of the ATXD signal is intermediate the values of the three backup sensor inputs AOTA-AOTC.

An additional input to the controller 42 from the analog backup means 36 is an analog temperature reference signal ATREF on a conductor 56. The ATREF signal is an analog voltage proportional to a desired turbine operating temperature, the value being set at a safe level for all possible operating conditions. In the embodiment of FIG. 1, the ATREF signal is shown as being generated by the analog backup means 36. However, the ATREF signal may come from any suitable analog voltage source for application to the controller. A value corresponding to the ATREF signal may also be stored in the controller instead of providing it by an external source. In any event the ATREF signal may be utilized in calculating the BTMTEM signal when the analog means is incorporated in the system. The ATREF signal is also used by the analog backup means 36 as will subsequently be described. The controller also receives a signal L26TC on a conductor 55 from the backup means 36 by way of a threshold detector 59. This signal is utilized by the controller 42 to determine when the system is on temperature control versus start or warm-up control.

The temperature analog backup means 36 generates two additional output signals, the temperature fuel control signal TFC onn a conductor 58 and a TRIP signal on a conductor 60. The TRIP signal is applied as a control signal to a fuel control valve 62 via a conductor 60.. It is utilized to shutoff fuel going to the turbine in the event of certain system failures to be subsequently described.

Reference is now made to FIG. 2 which shows, in schematic block diagram form, one embodiment of the analog backup means 36. This circuit is comprised of a mid-value selector 64 of known type for receiving the backup analog sensor inputs AOTA, AOTB and AOTC on conductors 50. The details of the mid-value selector 64 are described in U.S. Pat. No. 3,492,588, entitled "Median Selector For Redundant Signals", to M. P. Woodward, Jr., and assigned to the assignee of the present invention. The mid-value selector 64 provides the ATXD signal on conductor 54 to the controller 42. As previously mentioned, the ATXD signal, under normal sensor operating conditions, takes on a value mid the values of the signals applied to the mid-value selector. Further, the ATXD signal normally hhas a value approximately proportional to the average turbine exhaust temperature. The ATXD, BTMTEM and ATREF signals are all provided to a summer shown as a junction 66 where these signals are algebraically combined for input to a regulator amplifier 68, the output of which provides the TFC signal to the low value gate 34 and to a threshold detector 59. The threshold detector 59 generates the L26TC signal to the controller.

The ATXD signal is applied to the summing junction 66 through a resistor 70 and the BTMTEM signal is applied to the junction 66 via a resistor 72 The ATREF signal, the safe temperature reference, is provided to the junction 66 through a resistor 74 and to the controller on conductor 56. The ATREF signal is derived from a voltage source V and ground potential via a slider resistor 73.

Still referring to FIG. 2, the ATXD signal is also supplied to two conventional comparator amplifier latch circuits 76 and 78. Each of the comparator latch circuits has a voltage reference input from corresponding slider resistors 80 and 82, each connected between voltage source V and ground providing a voltage reference for comparison with their respective inputs. The comparator and latch circuit 76 has its input voltage from slider 80 adjusted to a specified voltage level whereby the comparator circuit is activated when the output signal ATXD from the selector 64 achieves a high value in excess of the specified voltage from slider 80. In a similar fashion, the slider 82 is adjusted to cause the comparator latch 78 to be activated when the ATXD signal achieves a low value in excess of the specified voltage setting of slider 82. Typically in the operation of the trip circuit of the analog backup means, when two or more of the sensors providing the AOTA through AOTC signals to the mid-value selector 64 are shorted, the signal ATXD will go to a low or negative value, activating the comparator latch 78. In a similar fashion, if two or more of the sensor input signals show an open condition, the output signal ATXD will go to a high voltage activating the comparator latch circuit 76. Each of the comparator outputs are tied back through their corresponding output conductors 84 and 86 to their latch circuits to cause each of the comparators to set to a latched state when activated. Whenever either one of the comparator latch circuits is activated, that circuit's output signal is passed through an OR gate 88 generating the TRIP signal on conductor 60 to activate the fuel control valve 62 (FIG. 1) shutting off fuel to the turbine. Further, the mid-value signal ATXD can exceed the threshold setting established by slider 80 should the turbine temperature become excessive. This will also cause a generation of the TRIP signal.

Whenever the turbine is shut down it will remain down until restarted through operator intervention. In order to reactivate the supply of fuel to the turbine, an operator must depress a reset switch 90 (FIG. 2) applying a suitable potential, such as a ground signal, to each of the latch circuits 76 and 78. This ground signal resets the activated comparator, thus deactivating the TRIP signal which opens the fuel control valve to allow fuel to flow to the turbine. Obviously, the supplying of fuel to the turbine is integrated with overall turbine start-up not a part of this invention.

Prior to proceeding with an overall operational description of the invention, it is considered advantageous to first reference FIG. 3 which illustrates an exemplary memory map for the storage of programs and data as stored in the memory of the programmable controller 42. The exemplary memory map is segmented into eight portions; (1) input data, (2) output data, (3) constants and masks, (4) master status program, (5) scratch, (6) temperature control program, (7) other control programs, and (8) sequencing program. The master status program may be considered as the executive program in the present embodiment in that it controls the reading of all information into the controller and the transfer of control information from the controller to the turbine. The master program is similar to a program described in the aforementioned U.S. Pat. No. 3,969,722 and identified therein as "Initialization and Standard Status Program" (ISSP).

Under control of the master status program, the temperature sensor input signals ATCX1-ATCX12 and AOTA-AOTC in FIG. 1 are read into the input data portion of the memory of FIG. 3 as temperature sensor conditions and temperature values. Also, the ATXD, ATREF annd L26TC signals are read into memory by the master status program. All of these input signals are stored in digital form in the input data portion of the memory subsequent to the performance of an analog to digital conversion thereon.

The output data portion of the memory of FIG. 3 contains calculated data in the form of a trim or fuel control signal for controlling the turbine which results from the execution of the temperature control program comprised of three subroutines TCC, TCA and PPR. As shown in FIG. 3, the temperature control program is utilized to calculate the temperature of the turbine and the value of the trim signal. The values calculated by the temperature control program are then stored in the output data portion of the memory for output through a digital to analog converter to the turbine.

The constants and masks portion of the memory has been included merely for completeness to show that conventionally the memory stores this type of data for use by the programs. For a similar reason, the scratch portion or working storage is shown to illustrate that the programs utilize a portion of memory into which data is written and read from during execution of the programs.

The other control programs portion of the memory in FIG. 3 is shown merely to indicate that other programs may reside in the memory; for example, programs to control the speed/load and the start-up circuits 38 and 40. The final portion of the memory, the sequencing program, is included for completeness to illustrate a program for simulating relay logic, etc. in electrical ladder diagram form. A program of this type is described in the aforementioned patent application and identified therein as an Application Program.

A detailed description of the master status and sequencing programs is not included in the present embodiment. A detailed description of programs of these types may be found in the aforementioned U.S. Pat. No. 3,969,722. A description of an analog to digital conversion program for converting the sensor inputs to digital values is not included since analog to digital conversion programs and hardware are well known in the art and such inclusion would add nothing to an overall understanding of the invention.

Reference is now made to FIG. 4 which is a high level flow chart showing the overall program of the present invention for controlling the gas turbine 10. The program is comprised of a master status program (MSP), a sequencing program (SEQ) and a temperature control program which includes a thermocouple calculation (TCC) subroutine, a temperature calculation (TCA) subroutine and a temperature trim (PPR) subroutine. The programs are shown as being sequentially executed in a loop such that, for each sweep cycle through the loop, the program returns to the MSP from the PPR.

Still referring to FIG. 4, when the control system of the present invention is first started up, the controller 42 effects an entry into the master status program (MSP) into an action block 92 whereby that program reads the input signals ATCX1-ATCX12, AOTA-AOTC, ATXD, ATREF and L26TC into the input data portion of the memory. After the input data is stored, the MSP enters block 94 and sends output signals BTMTEM and others, not shown, to the turbine from the output data portion of the memory. Obviously, on the first sweep cycle through the program there will be no data in the output data portion of memory. However, after the first or subsequent sweep(s) through the program, data will exist in the output data portion of the memory.

The program now proceeds to the sequencing program (SEQ), if implemented into the system, whereby the SEQ is executed to simulate a relay or logic controller by simulating relay coils, contacts, timers, etc. to develop output control signals for the turbine to actuate various indicator lights, solenoids, solenoid drivers, etc. not shown. The results of information derived from the execution of the SEQ program are stored in the output data portion of the memory for subsequent transfer to the turbine during block 94 of the MSP.

After execution of the SEQ program the temperature control program is entered at TCC, the thermocouple calculations subroutine of block 98. The TCC subroutine has one primary function and that is to determine the validity or reliability of the turbine temperature sensors by examining sensor words corresponding to the sensor temperature readings ATCX1-ATCX12 and AOTA-AOTC in the input data portion of the memory (FIG. 3). In the preferred embodiment, this determination is made by the TCC program examining a specified bit in each of the words corresponding to the sensor readings. This bit in each of the sensor words is referred to as a validity flag which is either set or reset during the analog to digital conversion of each of the sensor signals ATCX1-ATCX12 and AOTA-AOTC. It is well known in the art that overflow and underflow detection techniques are commonly used in analog-to-digital computation to determine whether the value of an analog signal is either above or below specified values. Another obvious way to determine the validity of sensor conditions would be to merely set arbitrary maximum values for the sensor words and to compare the actual value of each word with the predetermined values and set a flag of either of those predetermined values are exceeded.

The TCA subroutine is entered from TCC at an action block 100. The primary purpose of the TCA subroutine is to compute a common or average temperature from the number of valid readings previously calculated in the TCC subroutine. The TCA subroutine, upon its completion, effects an entry into the PPR subroutine shown by action block 102. The PPR subroutine, based on the results of the computed common or average temperature as established by the TCA subroutine, calculates the output trim signal BTMTEM.

Upon completion of the PPR subroutine, the program returns to the MSP and the above described sequences are repeated. As previously described, the BTMTEM signal is transferred to the turbine during the execution of block 94 of the MSP.

The overall operation of the TCC, TCA and PPR subroutines will now be described in connection with FIGS. 5 through 8 which illustrate in functional flow chart form the operational details of those subroutines.

Referring now to FIG. 5, the TCC subroutine is entered directly from block 96 of the sequencing program of FIG. 4. The TCC subroutine is a relatively simple subroutine, as shown by action block 104, wherein the program effects the setting of analog-to-digital (A/D) conversion error flags in a word in scratch or working storage designated TMERWD for each invalid or erroneous sensor reading. As indicated in block 104, one flag corresponds to each of the ATCX1-ATCX12 and AOTA-AOTC signals or conditions. This is accomplished by the TCC subroutine examining the previously mentioned flag bit in each of the sensor words to set or reset a corresponding error flag bit in location TMERWD in scratch storage in accordance with the status of each sensor flag bit. An error flag is set if its corresponding sensor word flag bit was determined to be invalid, otherwise the error flag bit is reset or cleared. An error flag bit which is set indicates that that flag bit's corresponding sensor (derived from a corresponding sensor word) is either open or shorted indicating an invalid reading.

The TCA subroutine is now entered into a program loop comprised of blocks 106, 108, 110 and 112. In this loop, a test is performed to see how many of the sensor input readings, if any, are invalid as defined by the set error flags in the thermocouple error word TMERWD. Upon each entry into this loop, a test is performed in block 106 on one bit of the TMERWD corresponding to one of the sensor input readings or words ATCX1-ATCX12, AOTA-AOTC to determine if the error flag bit in TMERWD is set or reset. Each time a flag bit in TMERWD is tested, if that bit is set, indicating that its corresponding sensor word is to be rejected, the program will exit block 106 through a YES branch entering into an action block 108. In block 108 a running record is kept of the number of rejects or invalid sensor readings. This recording of the rejected readings could be accomplished in a number of ways. In the present embodiment a program counter is employed. Prior to entry into block 108, this counter is loaded with a number representative to the number of sensor inputs (15 in the present embodiment). For each entry into block 108 the counter is decremented by one. Thus, the number in the counter always indicates the number of good readings remaining (i.e., the number of valid sensors existing in the system).

Upon each entry into block 106, if the flag bit being tested in the TMERWD is reset, indicating a valid reading, entry is made through the NO branch into block 110. Block 110 is an action block wherein the TCA subroutine finds the maximum (TMAX) temperature or value of the sensor readings, the minimum (TMIN) temperature or value of the sensor readings and the sum (TXSUM) temperature or value of the sensor readings. It will be noted that, since entry into block 110 is always via the NO branch of block 106, the values of TMAX, TMIN and TXSUM are always related to the unrejected or valid sensor readings. After each flag bit in location TMERWD has been tested, the program exits through a finished decision block 112 through a YES branch into a block 114.

In block 114 a common or average turbine temperature is calculated by storing a value representative of that temperature in a location TXAVGN, whereby the value of TXAVGN is representative of the value of TXSUM divided by the number (#) of good readings previously recorded in block 108. Upon entry into block 114, if all of the temperature sensor readings were found to be valid, the value placed in TXAVGN is the actual average turbine exhaust temperature. However, if one or more sensor readings were rejected, that is, found to be invalid, the value stored in TXAVGN is an average temperature of the valid temperature sensor readings.

Entry is now made into a decision block 116 wherein a test is performed to see if the difference between TMAX and TMIN is greater than a constant LTXSPD stored in the constants portion of the memory. This test checks to see if the spread between TMAX and TMIN is excessive. If the spread is not excessive, indicating that the temperature sensors are providing within limits readings, the program exits through a NO branch to point TCAC50 through a block 118. Block 118 is not an action block but has been included as a descriptor to indicate that the program is to use the actual average calculated temperature TXAVGN as the average temperature TXAVG. TXAVG is a location in scratch memory.

Still referring to FIG. 5, if the temperature spread is excessive, it indicates that a potential turbine combustion problem exists or possibly a temperature sensor is providing an erroneous reading and the program exits through the YES branch of block 116 to a point TCAB10 of FIG. 6. One potential problem which could exist is that something could possibly go wrong with a temperature sensor causing that sensor to generate either an artifically low or high reading not indicative of an invalid sensor condition. If this artificially low or high reading is factored into the overall common or average turbine exhaust temperature, it could cause the turbine to be artificially fired to generate more temperature than is required or to cut down fuel flow thus generating less temperature than is desired. (An overfiring of the turbine can cause serious damager to or degradation of its component parts whereas underfiring will cause a decrease in the performance of the turbine.) A second possible problem is that a clogged fuel nozzle in one of the combustion chambers could result in an indication of a cold spot or low temperature spot in the turbine exhaust. Should this condition occur, an out of limits reading will again cause the turbine to be overfired. (The reverse analogy could also be applied for hot spots in the turbine whereby a malfunctioning nozzle could be providing more fuel to the turbine than is required or specified). In view of the above problems, as detected by the temperature spread test in block 116, it is desirable to isolate definite or potential problems.

The flow chart of FIG. 6 shows the actions taken by the temperature control program to first of all identify the potential combustion problem areas in the turbine and to then recalculate the common or average temperature for use in controlling the turbine. Upon entry at point TCAB10 at block 120 the program initializes system parameters in preparation of testing for and isolating the combustion or sensor problems. This is accomplished by first establishing upper and lower temperature limits whereby a memory location UPPER (Upper Limit) is set equal to the value of TXAVGN, the calculated common or average temperature, plus the value specified by an upper temperature limit LTXDF1 in the constants portion of the memory. In a similar fashion, a lower temperature limit is established by setting a memory location LOWER equal to TXAVGN minus the value of a lower limit temperature constant LTXDF2. In block 120, LTXDF1 and LTXDF2 specify the maximum upper and lower temperature deviation limits of TXAVGN. Additionally, in preparation of recalculating the turbine temperature, the reject counter as previously established in block 108 of FIG. 5, is cleared.

Entry is now made into a decision block 122 wherein a test is again performed similar to that previously described for block 106 of FIG. 5 to see if any one of the sensor input readings is an invalid or reject reading. For each sensor reading determined invalid, a record rejects block 124 is entered, via the YES branch from block 122, wherein the rejected sensor readings are recorded as previously described in connection with block 108 of FIG. 5. Exit from block 124 is to a finished decision block 134. The program will continue to loop through the NO branch of block 134 back to block 122 until all of the sensor readings have been tested.

Reference is now made back to the NO branch of block 122 entering into a decision block 128. Block 128 is entered upon each pass into block 122 whenever the sensor reading being tested is not to be rejected. In block 128 a test is performed to see if the temperature specified by each of the temperature readings ATCX1-ATCX12 and AOTA-AOTC is greater than the lower limit previously established in block 120. Each reading that is not greater is recorded as a rejected reading in block 124. The rejection of a reading in block 128 is indicative of a combustion or temperature sensor problem. For each reading which is greater than the lower limit, a YES branch is taken into another decision block 130.

In a block 130 a similar test to block 128 is performed to see if any of the sensor readings are less than the upper limit as previously established in block 120. Any one of these readings which is out of limits will cause the program to take the NO branch and record that reading as a rejected sensor in block 132. Block 132 returns back to block 126. For each reading which is within limits, the program will take the YES branch entering into block 126. It will be noted that the record reject block entered from block 130 is shown as block 132; however, it represents the same counter as blocks 124 or 108. They are shown separately for simplicity and because they have different entry points to the subroutine.

In block 126 a new temperature sum is accumulated or calculated by setting location TXSUM equal to its present value plus the sum of the unrejected sensor readings (i.e., the number (#) of good readings left). Exit from block 126 is into the finished decision block 134 whereby a branch is taken through the NO exit back to block 122 until each of the sensor readings for inputs ATCX1-ATCX12 and AOTA-AOTC has been tested. The operation of this loop is somewhat similar to that previously described in connection with FIG. 5 for blocks 106, 108, 110 and 112. After all of the sensor readings in FIG. 6 have been tested, exit is made through the YES branch of block 134 to a point TCAC40 entering into a block 136.

In block 136, a new common or average temperature is calculated wherein location TXAVGN is now updated to the contents of the new or accumulated TXSUM divided by the number of new good readings established in blocks 124 and 132. The program now proceeds to a point TCA10 of FIG. 7 into a block 140. In block 140 the controller now tests for hot and cold spots in the turbine combustion chambers to identify potential combustion problems. This is accomplished by the program testing groups of temperature sensor readings corresponding to groups of adjacent sensors circumferentially disposed around the exhaust plenum. For example, the 15 sensors generating signals ATCX1-ATCX12 and AOTA-AOTC can be segmented into five groups of three sensors each. The program then checks each group of three sensors to see if the sensors in each group are providing valid within temperature limits readings. If any group shows a cold spot within the exhaust plenum, a cold spot flag TCACS is set in the output data portion of memory. If any group designates a hot spot reading, a hot spot flag TCAHS is set in the output data portion of memory. Though not shown in the flow charts of the present invention, the TCACS and TCAHS flags could be used by the program to generate an output signal during the running of the master status program to sound an alarm to notify an operator that either a cold or hot spot resides in the turbine. On completion of block 140, a block 142 is entered through a point TCAC50. It will be recalled that point TCAC50 is also entered from block 118 of FIG. 5 as previously described. Upon entry into block 142 the common or average temperature TXAVGN will be the actual average temperature as previously calculated in FIG. 5, or it will be the newly calculated common or average temperature as just described in connection with FIG. 6. In block 142 it is now desirable to test or check for the minimum number of temperature readings (i.e., ATCX1-ATCX12 and AOTA-AOTC) which are within limits. This is accomplished by the program subtracting the contents of a constant in memory from the contents of the counter specifying the number of good readings. As previously described the counter was set in blocks 124, 132 or 108. An excessive number of temperature readings out of limits flag, TCATN, will be either set or cleared (reset) in accordance with the results of this subtraction. TCATCN is set if an overflow condition occurs indicating that too many temperature readings are invalid (for example, the number of good readings is less than 8). The TCATCN is an abort flag which is tested in the PPR subroutine to cause the program to abort from computer control to analog backup control. As shown in block 142, the contents of location TCATCN are transferred to a location DTMABT in the output data portion of the memory during the SEQ routine. Upon completion of block 142, action block 144 is entered.

In block 144 a test or check is made to see if there has been an excessive change in the common or average temperature since the last temperature calculation. This is accomplished by testing to see if the contents of a memory constant LTXAVG is equal to the newly calculated temperature, TXAVGN, minus the old calculated temperature, TXAVG. For example, if the contents of LTXAVG are equal to 25 degrees F. and if that is the maximum allowable temperature change between any two entries into the temperature control program, an excessive temperature flag TCAAVG is set. If the contents of LTXAVG is less than 25 degrees F., then the TCAAVG flag is reset or cleared.

Entry is now made into block 146 wherein the common or average temperature is updated by setting location TXAVG (old temperature average) to the newly calculated average TXAVGN. A block 147 is now entered wherein a correction value in a scratch memory location KVAL1 is set equal to the difference between ATXD (the mid-value signal) and TXAVG (the present average temperature). Entry is now made to a decision block 148 of the PPR subroutine of FIG. 8 where a test is made to determine if the system should abort from computer control to the temperature analog backup means 36. This test is performed by testing to see if the abort bit DTMABT is set equal to a binary 1 (see block 142 of FIG. 7). If it is a binary 1, the program takes a YES branch into a block 150 wherein the trim signal BTMTEM is decremented a predetermined amount whereby the BTMTEM signal will ramp down toward zero in a step fashion for each pass through the temperature control program. As previously described, the BTMTEM signal is transferred to the analog backup means during execution of the MSP. The PPR then returns control to the MSP for entry thereto from block 150. It will be recalled, as previously described in connection with FIG. 2, whenever the trim signal BTMTEM is at zero, that signal is no longer a controlling factor and thus the output of the amplifier 68 is representative of the algebraic sum of the mid-value signal ATXD and the analog reference ATREF. It is in this manner, when the computer output is aborted by the BTMTEM signal going to zero, that the analog backup means takes over control of the turbine to provide the temperature fuel control signal TFC on conductor 58 to the low value gate 34 to control fuel flow to the turbine. It is also significant to point out that most industrial control computers of the type contemplated by the present invention usually encompass test routines to check for the validity of computer operation (e.g., memory parity, analog input/output accuracy, etc.). Additionally, some type of "watch dog" timer is sometimes included. This timer is periodically interrogated by the program to insure that the computer is operating properly. The detection of any computer failure may also be used to cause the BTMTEM signal to be ramped to zero in a manner similar to that described in block 150.

Again referring to block 148, if the abort flag DTMABT is not set, the program will exit through the NO branch entering into an action block 152. In this block a delta temperature bias reference DELR1 is set up in memory to a value equal to the temperature base (TBASE) minus the value of the analog temperature reference (ATREF). TBASE is a turbine temperature set point normally operator specified. The DELR1 reference is used as one factor in developing the correct value for the BTMTEM signal. It has application because of the expected differences between the values of TBASE (e.g., 850° F.) and ATREF (e.g., 800° F.) and the difference value (DELR1) between TBASE and ATREF serve as the desired temperature reference or bias for summation with temperature variables to be described to develop the trim signal BTMTEM.

It will be recalled that ATREF always has a value which is safe for all turbine temperature operating conditions. Additionally, in block 152, a memory location KVALR1 is set equal to the correction value of KVAL1 previously established in block 147 of FIG. 7 whereby KVAL1 is a ramped value which causes the BTMTEM signal to change as an integrated signal to thereby precisely control the operation of the turbine so that radical changes in fuel flow do not occur.

The program now proceeds to a block 155 wherein a test is performed to see if the system is on temperature control (TEMP CONTROL?) by testing to see if a flag bit L26TC in the scratch portion of memory is a binary 1. It will be recalled that the signal L26TC on conductor 55 of FIGS. 1 and 2 was read into the controller by the MSP. Referring to the threshold detector 59 of FIG. 2, the L26TC will go to a binary 1 when the value of the TFC signal reaches a predetermined magnitude indicating that the turbine has cycled through a suitable warm-up period during system start-up. When the L26TC signal goes from a binary 0 to a binary 1, that binary 1 is stored in location L26TC and indicates to the program that the turbine is at the proper temperature to now go on complete temperature control.

If the system is not on temperature control (L26TC=1), the program enters into a block 154 via a NO branch of block 155. Assume, however, that the system is on temperature control, and thus the program exits block 155 through a YES branch into a block 157. In block 157 the difference between the base temperature (TBASE) and the highly accurate calculated average temperature (TXAVG) is calculated. A value proportional to this difference is stored in a location KVAL2 in the scratch or working storage portion of memory.

A block 159 is now entered and the value of KVAL2 is tested to see if it is less than or equal to a predetermined constant shown as zero (0). If there is a difference between KVAL2 and the constant zero, a location KVALR2 in scratch memory is appropriately changed by incrementing or decrementing its value. As will subsequently be described, KVALR2 is a second correction factor used in the equation for developing the temperature trim signal to control fuel flow to the turbine. The purpose of changing KVALR2 is to narrow the gap or difference between the TBASE and the TXAVG. KVALR2 is incremented or decremented by a vernier constant such as one or some other value. In this manner KVALR2 has a very slow ramp or rate of change so that the difference between the TBASE and the TXAVG is gradually tuned out to thus compensate for the system offset and drift errors introduced by the aforementioned electronic elements. With the difference tuned out, the total system error is reduced to that of only the temperature sensing portion of the system since there will be no discrepancy between the base temperature and the calculated average temperature.

The value of KVALR2 also has maximum positive and negative limits as specified by two locations in the constants portion of memory shown in block 159 as ± LKVAL2.

Upon completion of the test in block 159, a decision block 161 is entered wherein a test is performed to see if the absolute value of KVALR2 has exceeded either the positive or negative limits specified by locations ± LKVAL2. If KVALR2 has exceeded either of the limits, the program enters a block 163 through a YES branch. In block 163 an alarm flag in the output data portion of memory is set. This flag may be used to turn on an alarm when the master status program (MSP) sends the output data from the output data portion of the memory. When used to turn on an alarm, it can signal an operator that some malfunction exists causing an error so large that it cannot be calibrated out or compensated for by the adjustment of KVALR2. Such errors could exist in the controller output digital to analog converter, various control amplifiers, summing resistor tolerances, etc.

Once the alarm flag is set, the program enters into a block 154, the purpose of which will subsequently be described. Let it now be assumed that KVALR2 is within the limits of ± LKVAL2. As a result a NO branch from block 161 is taken into the action block 154.

In block 154, the value of the temperature trim signal BTMTEM is calculated by storing in a location BTMTEM in the output data portion of the memory the algebraic sum of the contents of locations DELR1, KVALR1 and KVALR2. The program will now proceed back to the master status program MSP wherein BTMTEM is transferred to the analog backup means.

For completeness of disclosure a copy of an assembly program listing of the temperature control program for operation in the aforementioned "Directo-Matic" logic controller is annexed hereto as "Appendix A" forming a part of this specification. It is recognized that system software (i.e., programs) are sometimes characterized with minor faults known as bugs which sometimes take long periods of time to detect and/or diagnose. Ordinarily the correction of such faults is within the skill and control of system programmers. Accordingly, the appended listing may be expected to contain some faults of this kind but all such faults which have been detected required only a programmer's skill for correction.

Reference is now made to FIGS. 9, 9a and 9b. FIG. 9 is a legendary drawing showing the interrelationship between FIGS. 9a and 9b. FIGS. 9a and 9b illustrate, as a second embodiment of the invention, an analog controller for monitoring the sensor signals ATCX1 through AOTC on conductors 48 from the turbine sensors 26. In this embodiment the analog controller of FIGS. 9a and 9b replace the programmable controller 42 of FIG. 1. In this embodiment the temperature analog backup means 36 is utilized in controlling the operation of fuel flow to the turbine. This is illustrated in FIG. 9b where the TFC signal from the analog means 36 is applied directly to the low value gate 34 as previously described in connection with FIG. 1.

Reference is now made to FIG. 9a. As shown there, each of the ATCX1 through AOTC sensor signals on conductors 48 is applied to a corresponding comparator 158 shown within dashed lines. While only two pairs or groups of comparators 158 are shown, it is to be understood that there is one comparator for each of the sensor signals. Each of two comparator circuits in the comparators 158 receive a reference signal from a corresponding reference source. For example, a comparator circuit 160, designated COMP HI, in each of the comparators 158 receives a high reference signal RHI on a conductor 162 from a high reference source 164 REF HI. In a similar fashion, a low comparator circuit 166, COMP LO, in each of the comparators 158 receives a low reference value signal RLO on a conductor 168 from a low reference signal source 170 designated REF LO.

In the embodiment of FIGS. 9a and 9b, the validity of the sensors 26 (FIG. 1) is accomplished by the comparators 158 comparing the values of the ATCX1 through AOTC signals with the RLO and RHI signals. In normal operation, the values of the ATCX1 through AOTC signals are within the limits prescribed by the RLO and RHI signals, and the output of each of the comparator circuits 160 and 166 is a binary 0. The outputs from each pair of comparators is applied to a corresponding OR gate 172 via respective conductors 174 and 176. So long as the signals on conductors 174 and 176 to an OR gate 172 are both binary 0's, the output of that OR gate is a binary 0 on a corresponding conductor 178. The output of each of the OR gates 172 is connected to a corresponding one of a plurality of switches in a sensor select switches -1 circuit 180. Two of these switches 182 and 184 are shown. Additionally, each of the switches receives the sensor signal corresponding to each switch's comparator. For example, switch 182 receives the ATCX1 sensor signal feeding that switch's corresponding comparator and switch 184 receives the AOTC signal, etc. Each of the plurality of switches (e.g., 182 and 184) has an inverter input terminal and each is enabled to pass its sensor input signal to an averager -1 circuit 192 via a plurality of corresponding conductors when that switch's input on conductor 178 is a binary 0. When the signal on conductor 178 is a binary 1, the switch receiving that signal is disabled to inhibit passage of its input sensor signal.

As an example of how the comparators 158 and select switches 180 operate, let it be assumed that the ATCX1 signal exceeds either the low or high limit as specified by the RLO and RHI signals. In this situation, the appropriate one of the comparators 160 and 166 will selectively determine the validity of the sensor generating the ATCX1 signal by generating a binary 1 signal on one of the conductors 174 or 176 as determined by the activated comparator. When this occurs, the binary 1 signal enables OR gate 172 to apply a binary 1 disable signal to switch 182, thus preventing the ATCX1 signal from being passed through the select switches 180 to the averager 192.

The averager 192 generates an output signal TXAVGN on a conductor 194. The value of the TXAVGN signal is always proportional to the average of the values of the ATXC1 through AOTC signals passed by the select switches 180. The TXAVGN signal is applied to the positive (+) terminal of an algebraic summer 196. The summer 196 also receives an input signal on a conductor 192 from an LTXDF2 reference source 200. It will be recalled, from the description of the first embodiment, that the LTXDF2 reference source specifies a lower limit temperature deviation of the average of the turbine temperature as specified by the TXAVGN signal.

The output of the summer 196 provides a low turbine temperature reference (NEW RLO) having the value TXAVGNLTXDF2 on a conductor 202. The NEW RLO signal is applied to each one of a plurality of comparators 204 designated COMP NEW LO. It will be noted that there is a corresponding comparator 204 associated with each of the ATCX1 to AOTC sensor signals with only two being shown. The purpose of each of the comparators 204 is to compare the NEW RLO value with each comparator's corresponding sensor input signal to determine if a potential combustion problem or abnormal sensor reading exists in the turbine in the area of the sensor generating the corresponding sensor signal. In normal operation when each of the ATCX1 through AOTC signals are within the temperature deviation limits as specified by the NEW RLO signal, the output of each of the comparators 204 will be a binary 0. The output of each of the comparators 204 is applied via a corresponding one of a plurality of conductors 207 to a sensor select switches -2 circuit 208. The select switches 208 are similar to the select switches 180 and operate in the same manner as previously described for the latter switches. It will be noted that a switch exists in the select switches 208 for each of the sensor signals ATCX1 through AOTC. For example, the ATCX1 signal is applied via a conductor 210 to the top switch 212 of select switches 208 and the AOTC signal is applied to a switch 214, the bottom switch, via a conductor 216. So long as the outputs of each of the comparators 204 are binary 0's, their corresponding switches 212 through 214 will be enabled to pass the ATCX1 through AOTC signals on a plurality of conductors 218 as signals ATCX1' through AOTC' to a second averager -2 circuit 220.

The averager 220 is the same type as the previously described averager 192. The averager 220 generates an output signal TXAVG on a conductor 222. The TXAVG signal has a value proportional to the average calculated turbine temperature.

Reference is now made back to the comparator circuits 204 of FIG. 9b. If any one of the ATCX1 through AOTC signals is greater than the low reference limit signal NEW RLO, the output of the comparator(s) receiving the out of limits sensor signal will go to a binary 1. This binary 1 signal, via conductors 207, will disable the switch receiving the signal to prevent the passage of its sensor input signal to the select switches 208. Thus, it can be seen that the output of the averager 220 always generates a TXAVG signal having a value representative of the average of the signals passed by the select switches 208.

The TXAVG signal on conductor 222 is applied to a negative (-) input terminal of a second algebraic summer 224. Summer 224 also receives a temperature base reference signal TBASE on a conductor 226 from a TBASE REF source 228. The output of the summer 224 is the previously described KVAL2 signal on a conductor 230 having a value of TBASE minus TXAVG. The KVAL2 signal is applied to a conventional deadband threshold detector amplifier circuit 232 which generates either a zero, positive, or negative output signal in accordance with the value of the KVAL2 signal. If the KVAL2 signal is zero, the output of detector 232 is zero. However, if the KVAL2 signal is above zero, the output of detector 232 will go to a positive value. The reverse is true if the KVAL2 signal goes below zero. Then, the output of detector 232 goes negative.

The output of detector 232 is applied via a conductor 233 to a conventional integrator amplifier 235 which integrates the input signal at a vernier constant slow rate over a period of time to generate the aforedescribed KVALR2 correction signal on a conductor 237. The KVALR2 signal is applied to two conventional limit comparator circuits 239 and 241 and compared with positive and negative reference limit signals from +LKVAL2 and -LKVAL2 limit reference sources 243 and 245, respectively. The output of each of the comparators 239 and 241 is applied to respective inputs of an OR gate 247. When the KVALR2 signal is within the limits of the ±LKVAL2 reference signals, the output of the comparators are both binary 0 keeping OR gate 247 disabled. If a system offset or drift error occurs which is outside of the limits prescribed by the ± LKVAL2 reference signals, the appropriate one of the comparators 239 or 241 will detect the out of limits condition and apply a binary 1 signal to OR gate 247 enabling the latter to generate a binary 1 signal to turn on an alarm not shown. As previously described, the alarm can be used to indicate to an operator an out of limits condition representative of the possible failure of one of the aforementioned elements.

Still referring to FIG. 9b, the calculated average temperature signal TXAVG is applied on a conductor 249 to a negative (-) input terminal of a summer 251. Summer 251 also receives the ATXD mid-value signal at a positive (+) input terminal from the analog backup means 36 via conductor 54. The TXAVG and ATXD signals are algebraically combined in summer 251 to generate the previously described correction signal KVAL1 on a conductor 253. The KVAL1 signal is proportional to the difference between the calculated average temperature (TXAVG) and the mid-value signal (ATXD). It will be recalled that the ATXD signal is derived from the AOTA-AOTC signals and that the ATXD signal has a value substantially equal to the average of the values of the AOTA-AOTC signals. Thus, the ATXD signal is substantially proportional to the average turbine temperature derived from the three sensors generating the AOTA-AOTC signals.

The purpose of combining the ATXD and TXAVG signals is to compensate or correct for the expected differences which exist between the highly accurate TXAVG signal (calculated from a possible 15 sensors) and the less accurate ATXD signal derived from three sensors to generate the correction value signal KVAL1.

The KVAL1 signal conductor 253 is applied to a positive (+) terminal of another summer 255 having an output connected to the input of a conventional integrator amplifier circuit 257. The output of amplifier 257 is the aforementioned KVALR1 signal on a conductor 259 applied to a negative (-) input terminal of the summer 255 and an output summer 261.

The KVALR1 signal is fed back to summer 255 and combined with the KVAL1 signal to provide damping so that radical changes in the KVAL1 signal are not reflected in the KVALR1 signal. It is significant to note that the KVALR1 signal is a much faster changing signal than the KVALR2 signal and serves as the primary fuel control parameter for the development of the BTMTEM signal on conductor 52.

The KVALR1 and KVALR2 signals are algebraically combined in summer 261. Summer 261 also receives the aforementioned delta temperature bias reference signal DELR1 from a summer 263 via a conductor 265. The DELR1 signal has a value proportional to the difference between the TBASE on a conductor 267 and the ATREF on a conductor 269 from the analog means 36. The output of summer 261 is the BTMTEM signal on conductor 52 having a value proportional to the algebraic sum of the DELR1, KVALR1 and KVALR2 signals. By referring to FIG. 2, it can be seen how the BTMTEM signal is combined with the ATXD and ATREF signals in the analog means to generate the temperature fuel control signal TFC to the low value gate 34 of FIG. 1.

Reference is now made back to the output of each of the comparators 204 of FIG. 9b. It will be noted that each of those comparators is connected via a corresponding conductor 234 to the input of a # good readings <8 decode circuit 236. The circuit 236 is a conventional decode network which monitors the binary signals on conductors 234 (15 in the preferred embodiments) to determine if the number of good readings as determined by the output of the comparators 204 is less than 8. So long as 8 or more of the signals on conductors 234 are binary 0's, the output of the decode network 236 is a binary 0 on a conductor 238. The output signal on conductor 238 from decode 236 is designated as a RESET INT signal to reset the two integrators 235 and 257 when that signal goes to a binary 1. If less than 8 of the sensors 26 are generating out of limits or invalid readings, the decode 236 generates a binary 1 output signal to reset the integrators 235 and 257. This resetting causes the output signal BTMTEM of the summer 261 to be ramped to zero, thus allowing the analog backup means 36 to operate the turbine at reduced safe temperature in a fail safe mode as previously described.

Reference is now made to FIGS. 10 and 11 which are legendary drawings showing the interrelationships of FIGS. 10a through 10e and 11a through 11d, respectively. FIGS. 10a through 10e comprise a third embodiment of the invention showing in detailed block diagram form a digital logic controller for calculating the value of the temperature fuel control signal TFC. FIGS. 11a through 11d collectively illustrate a timing diagram of the relationships between the various signals generated by the digital controller of FIGS. 10a-10e. In the ensuing description of the controller, the timing diagram of FIGS. 11a-11d will be referenced.

Reference is first made to FIG. 10a which shows a clock generator 250 providing a sequentially occurring clock signal CLK on a conductor 252 for application to various logic circuits and elements in the controller. The CLK signal is shown at the top of FIGS. 11a and 11b. In FIG. 10c there is shown a manual reset (MAN RESET) switch 254 having one terminal connected to a voltage source V. When the MAN RESET switch is placed in the closed position, a system reset signal is generated on a conductor 256. This latter signal is applied to the various registers, counters, flip-flops and elements in the controller to initialize or reset the system at start-up. The MAN RESET switch also applies the system reset signal as one input to an OR gate 258, enabling that OR gate to apply a binary 1 trigger pulse to a delay one-shot multivibrator 260. The delay one-shot generates a present output pulse on a conductor 262 to preset two counters designated # Good Reading Counters 1 and 2 circuits 264 and 266 in FIGS. 10c and 10b, respectively. Counters 264 and 266 are the equivalent to those counters previously described in connection with FIGS. 5 and 6 designated by blocks 108, 124 and 132. At the time of system initialization, each of these counters is preset to a count equal to the number of sensor input signals ATCX1 through AOTC (15 in the present embodiment).

Prior to proceeding with an operational description of the controller of FIGS. 10a-10e, it is considered first advantageous to briefly describe the operation of a two-bit delay counter 268, a scanner counter 270, a scanner 272 and an analog-to-digital converter 274. Each of these elements is illustrated in FIG. 10a. The two-bit delay counter 268 is a conventional binary counter having set (S), reset (R) and trigger (T) input terminals for generating three sequentially delayed output signals DLC0, DLC1 and DLC2. The timing relationships of these signals are shown in FIGS. 11a and 11b. The DLC2 signal from counter 268 is applied to the scanner counter 270 to an INC input terminal to cause that counter to be incremented by 1 each time the analog-to-digital converter 274 starts a conversion of one of the sensor input signals ATCX1 through AOTC on conductors 48. As can be seen, the scanner counter 270 provides a plurality of scanner count output signals SC00-SC14 on conductors 276 to the conventional scanner 272 to cause that element to sequentially scan or select the sensor input signals in accordance with each count in the scanner counter 270.

At the start of each analog-to-digital conversion, the DLC2 signal is applied to the A/D converter 274 as a reset/start A/D signal. The reset/start A/D signal initializes the A/D converter 274 to start its operation and simultaneously causes an A/D COMP signal on a conductor 278 to go to a binary 0 as shown in FIG. 11a. When the A/D COMP signal goes to a binary 0, that signal is applied as a binary 1 through an inverter input terminal of an AND gate 280. AND gate 280 is enabled when the first CLK signal, following the A/D COMP signal, is applied to that gate to provide a reset signal to the R terminal of counter 268. This effects the generation of the DLC0 signal at the time shown in FIGS. 11a and 11b. At the completion of the analog-to-digital conversion, the A/D COMP signal goes to a binary 1 applying a set input signal to the S input terminal of the delay counter 268.

As shown in FIGS. 10a, 11a and 11b, the delay counter 268 is now enabled by the A/D COMP binary 1 signal to generate the DLC1 and DLC2 signals in response to the two CLK signals applied to the T terminal of counter 268. As previously described, each time the DLC2 signal is generated the scanner counter 270 is incremented to cause the scanner to sequentially select the next sensor input signal and simultaneously reset and start the A/D converter 274. The functions just described are repeated for each A/D conversion which the controller performs on the various sensor input signals ATCX1 through AOTC.

Let it now be assumed that the system has been initialized and that the controller has just completed an analog-to-digital conversion on the ATCX1 signal. As illustrated in FIG. 11a, the scanner counter 270 is at an all binary 0 state, thus generating a binary 1 SC00 signal to cause the scanner 272 to select the ATCXl signal as shown in FIG. 10a. The A/D converter 274 is also capable of generating an output signal OV+UV on a conductor 282. The A/D converter 274, which is conventional by design, will generate a binary 1 output signal on conductor 282 when it detects either an overflow or an underflow condition. The OV+UV signal on conductor 282 is representative of such a detected overflow or underflow condition. In normal operation, if the sensor providing the selected ATCXl through AOTC signal is not opened or shorted, the OV+UV signal will remain at a binary 0. As shown in FIG. 10a, the OV+UV signal is applied on conductor 282 to an inverter input terminal of an AND gate 284 in conjunction with the A/D COMP signal. Thus, it can be seen, if the A/D converter 274 does not detect an overflow or underflow condition, AND gate 284 will be enabled to provide a binary 1 set input signal to an S terminal of a Good F/F #1, 286. The timing for the setting and resetting of flip-flop 286 is illustrated in FIGS. 11a and 11b in dashed line form. The dashed lines indicate that flip-flop 286 will set if there is no overflow or underflow condition, otherwise it will remain reset. It is to be noted that flip-flop 286 is always reset by the DLC2 signal from the delay counter 268 each time an analog-to-digital conversion is started.

Let it now be assumed that the Good F/F #1, 286, is in a set state. Thus, that flip-flop generates a binary 1 output signal on a 1 terminal which is applied via a conductor 288 to an enable EN input terminal of a Switch #1, 290, and as one input to an AND gate 292. Switch 290 is also receiving the converted sensor data in digital form from the A/D converter on a plurality of conductors 294. With switch 290 now enabled, the sensor data is passed through switch 290 to the input of a conventional parallel type adder designated Adder #1, 296. Adder 296 is now enabled by an ADD signal on a conductor 298 from AND gate 292 which is enabled by the A/D COMP signal on conductor 278 from the A/D converter 274. The generation of the ADD signal on conductor 298 is shown in FIGS. 11a-11b in dashed line form indicating that that signal will be generated only when flip-flop 286 is set. If flip-flop 286 is not set, indicating an invalid condition of the sensor being monitored by the A/D converter, the adder 296 is not enabled to perform its add function.

Let it now be assumed that the adder 296 is enabled to perform its add function. As shown in FIG. 10a, the adder 296 provides output signals on a plurality of conductors 300 to a conventional accumulator register 302 designated ACC #1 and receives input signals via a plurality of conductors 304 from the accumulator 302. Thus, it can be seen that, each time adder 296 performs an add function, the contents of the accumulator 302 are added to the sensor data input signals from switch 290 and the sum of the addition is provided back to the accumulator 302.

Reference is now made to FIGS. 10a and 10b and the timing diagram of FIGS. 11a and 11b. It will be noted that the ADD signal on conductor 298 is also provided to an enable EN input terminal of a conventional input multiplexer circuit 306. The multiplexer 306 also receives the sensor data from the output of switch 290. When the ADD signal is generated by gate 292, multiplexer 306 is enabled to pass the sensor data via a plurality of conductors 308 into the top register of a plurality of sensor data registers 310. Sensor data registers 310 preferably are comprised of a converter pushdown type stack, whereby the information entering into the top register is sequentially shifted down into the next register upon activation of the sensor data registers by the DLCl signal.

By referring to FIGS. 11a and 11b, it can be seen that while the sum is being accumulated in accumulator 302, the sensor data is placed into the sensor data registers 310 by the DLCl signal gating registers 310. Also at this time, it will be noted that the DLCl signal is applied to a set S input terminal of a sensor data operation flip-flop SDO F/F 312. Flip-flop 312 is set at CLCl time to generate an SDO1 binary 1 signal on a conductor 314 as shown in the timing diagram of FIGS. 11c and 11d. The SDO1 signal is applied to an enable EN input terminal of a conventional output multiplexer circuit 316. The last register stage of the sensor data registers 310 provides sensor data via a plurality of conductors 318 to the input of the output multiplexer 316. Thus, it can be seen that multiplexer 316 is enabled by the SDO1 signal to pass the sensor data from the final output stage of registers 310 on a plurality of conductors 320.

Reference is now made back to the sensor data registers 310 of FIG. 10b. Preferably, registers 310 contain a number of registers equal to the number of sensor input signals ATCX1 through AOTC (15 in the present embodiment). As such, it can be seen that it will take 15 A/D conversions before the sensor data registers 310 are each filled with digital data representative of the temperature values of their corresponding ATCX1 through AOTC signals. Further, it should be noted that, at the beginning of each new scan by scanner 272, the ATCX1 sensor signal is the first signal to be entered into the sensor data registers 310. Thus, after 15A/D conversions, the ATCX1 signal contents will be in the final output register of registers 310.

it is also significant to note at this time, that any sensor input signal which is detected as either opened or shorted, indicating an invalid sensor condition, prevents generation of the ADD signal on conductor 298. As a result, the input multiplexer 306 is not enabled and binary 0's are thus passed into the top register of the sensor data registers 310 for that particular sensor. As a result, registers 310 always contain representations of the values of only those sensor readings ATCX1 through AOTC which are considered to be valid. All invalid conditions are held in registers corresponding to their sensor signals as binary 0's.

To understand the operation of the logic shown in FIGS. 10b, 10d and 10e, it is now considered advantageous to refer back to the A/D converter 274 of FIG. 10a. The A/D COMP and OV+UV output signals from converter 274 are applied via conductors 278 and 282 to an AND gate 322. The output of AND gate 322 is a decrement DEC signal which is applied to the # Good Readings counter #1, 264, to decrement that counter each time the A/D converter detects either an overflow or underflow condition resulting from an invalid sensor reading. Since counter 264 is preset at the beginning of each scan, it always contains a count representative of the number of good or unrejected readings taken by the A/D converter 274. Signals representative of the number of good readings in counter 264 are applied via a plurality of conductors 324 to a conventional divider circuit designated Divider #1, 326. Divider 326 also receives signals representative of the accumulated sum in accumulator 302 via a plurality of conductors 328. When the divider 326 is activated to perform a divide function, it divides the number of good readings on conductors 324 into the accumulated sum from accumulator 302. The result of this division is the calculated average temperature shown as a TXAVGN signal from the output of divider 326 on a plurality of conductors 330.

The divide operation is enabled by a start DVD #1 signal applied to an EN input terminal of divider 326 via a conductor 332 from a 1 output terminal of a START DVD F/F #1, 334. The flip-flop 334 is set by a binary 1 START DVD signal from the output of an AND gate 336. By referring to FIGS. 11c and 11d, it can be seen that AND gate 336 is enabled when the two-bit delay counter 268 generates the DLC2 signal and a scanner count decode generates a binary 1 scanner count decode zero signal SCD0. The decode network 338 receives the SC00 through SC14 signals via a plurality of conductors 340 from the scanner counter 270. As shown in FIGS. 11b and 11d, whenever the scanner counter 270 achieves a count of zero, the output of the scanner count decode 338 generates the SCD0 signal to enable AND gate 336 at DLC2 time, thus generating the START DVD signal. The first CLK signal applied to a T terminal of flip-flop 334 following the generation of the START DVD signal, sets flip-flop 334 generating a binary START DVD1 signal on a conductor 332. When the START DVD1 signal is applied to divider 326, the # of good readings signals on conductors 324, and the accumulated sum on conductors 328 are clocked into the divider 326 where they are stored in registers internal thereto and not shown. The CLK signal applied to divider 326 controls the timing of that circuit in performing the divide function.

The divider, upon the completion of the divide operation, generates a DVD1 COMP binary signal on a conductor 342. The DVD1 COMP signal is applied to a reset R input terminal of flip-flop 334 resetting that flip-flop. The length of time for performing the divide can be variable, so long as the divide operation is completed before the next SC00 signal is generated. Also, the DVD1 COMP signal is applied to a TXAVGN register 344 which receives the calculated average temperature TXAVGN at the time the DVD1 COMP signal is generated. Additionally, the DVD1 COMP signal is applied to a conventional adder circuit 346 which algebraically adds the calculated average temperature TXAVGN to a lower limit temperature deviation limit or value LTXDF2 from an LTXDF2 reference source 348. The DVD1 COMP signal is a binary 1 pulse having a duration of approximately one CLK period, which is sufficiently long to allow the adder 346 to perform the addition of the LTXDF2 and TXAVGN signals.

When the DVD1 COMP signal returns to a binary 0, that signal is inverted to a binary 1 to a DVD1 COMP signal through an inverter 350 of FIG. 10c. When the DVD1 COMP signal goes to a binary 1, the contents of the ADDER 346 are clocked into a lower limit register 352 via a plurality of conductors 354. The contents of the lower limit register represent the maximum lower limit of the calculated average temperature. This lower limit is provided as an output signal reference on a plurality of conductors 356 from Register 352. The lower limit value on conductors 356 is applied to a conventional subtracter 358. Subtracter 358 is also receiving the sensor data output from the multiplexer 316 via conductors 320 as sensor data' signals.

It is to be noted at this time that the divide operation just described is performed only once during each complete cycle of the scanner 272. Each time the scanner count decode 338 of FIG. 10c denotes a count of zero (SCD0=binary 1), signifying the beginning of a new scan, the START DVD signal is generated. The START DVD signal is also applied via OR gate 258 to the delay one-shot 260 which generates a delayed output pulse to preset the # Good Readings counter #1, 264, at the time noted in FIG. 11c. Thus, counter 264 is always preset to the proper value at the beginning of each new scan.

Additionally, the lower limit register 352 always contains the lower limit value resulting from each divide operation. As such, the contents of register 352 can be changed only once during each scan of the sensor input signals ATCX1 through AOTC.

Reference is now made back to FIG. 10d to the subtracter 358. The purpose of the subtracter 358 is to subtract each of the ATCXl through AOTC readings from the lower limit reference just described to detect for potential combustion problems in the turbine. The operation of the subtracter 358 can best be understood by first referring to a zero decode 360 receiving the sensor data' signals on conductors 320 from the multiplexer 316. The decode 360 is conventional in design and continuously monitors the output of multiplexer 316 to see if any of the ATCX1 through AOTC readings are all binary 0's, indicating a rejected temperature sensor. Under normal conditions, that is when a sensor has not been rejected, the input signals on conductors 320 to the decode 360 will always be representative of some value other than zero. So long as the decode 360 is decoding a zero input on conductors 320, a binary 1 zero output signal on a conductor 362 will be present. The zero, signal, in conjunction with the previously mentioned SDO1 signal from flip-flop 312, is applied to an AND gate 364. As shown in FIGS. 11c and 11d, AND gate 364 generates a binary 1 subtract enable signal SUBEN on a conductor 365 each time the SDO1 signal goes to a binary 1 and the zero decode output signal is a binary 1. When the SUBEN signal goes to a binary 1, that signal is applied as an enable input to an EN input terminal of substracter 358 to effect the previously mentioned subtraction. If the subtraction of any one of the ATCX1 through AOTC signals from the lower limit results in an overflow condition in the subtracter, a binary 1 out of limits signal is generated on a conductor 366. The out of limits signal is combined in an AND gate 368 with the DLCO signal to enable that AND gate during an out of limits condition to thus generate a binary 1 signal for application to an OR gate 370. OR gate 370 is enabled each time an out of limits condition occurs to apply a binary 1 decrement DEC' signal to a second # Good Readings Counter #2, 266, to decrement that counter by 1 for each out of limits condition detected by the subtracter 358. The DEC' signal appears on a conductor 371.

Reference is now made back to the output of AND gate 364 of FIG. 10b. As previously described, under normal operating conditions the output of that AND gate is a binary 1 to generate the SUBEN signal. It should be noted that the SUBEN signal is inverted through an inverter input terminal applied to an AND gate 372, also receiving the SDO1 and DLCO signals. Each time the zero decode 360 decodes a binary 0 value from multiplexer 316, AND gate 364 is disabled to provide a binary 0 signal on the SUBEN line. This signal is inverted to a binary 1 to thus enable AND gate 372 at the time indicated by the generation of the DEC' signal in FIGS. 11c and 11d. In other words, any time a zero is decoded by decode 360, indicating that the sensor corresponding to that reading being decoded at that time is a rejected sensor, counter 266 is decremented by 1. This decrementing, in this instance, is provided by the output of AND gate 372 enabling OR gate 370 to generate the DEC' binary 1 signal. It can now be seen that counter 266 always contains a count representative of the number of good readings determined by the rejection of the invalid sensors during the A/D conversion process and by those rejected sensors providing out of limits readings during the subtraction process by subtractor 358.

Reference is now made to FIG. 10b to an inverter input terminal of an AND gate 374 receiving the out of limits signal on conductor 366. AND gate 374 also receives the DLCO and SDO1 signals At the completion of a subtraction operation, if there is no out of limits condition, the binary 0 signal on conductor 366 is inverted to a binary 1 through the inverter of AND gate 374 to thus allow that gate to be enabled upon the occurrence of the DLCO and SDO1 signals to set a Good F/F #2, 376, at the time shown in FIGS. 11a-11d. Setting flip-flop 376 provides a binary 1 output signal ADD' on a conductor 378 to an enable EN input terminal of a switch #2, 380, and to an enabled EN input terminal of an ADDER #2, 382. The switch 380 also receives the sensor data' signals via conductors 320 and passes that data into adder 382 upon the occurrence of the ADD' signal. The operation of switch 380 and adder 382 is the same as previously described for switch 290 and adder 296. Adder 382 provides data via a plurality of conductors 384 to an accumulator ACC #2, 386, and receives data therefrom via a plurality of conductors 388. Accumulator 386 operates in the same fashion as previously described for accumulator 302 to provide SUM' data signals to a divider #2, 394, on a plurality of conductors 390. The Good F/F #2, 376, is reset by the SDO and DLC1 signals applied to an AND gate 392 generating a binary 1 signal to a reset R input terminal of flip-flop 376 to terminate the operation of ADDER 382 at the time shown in FIGS. 11c and 11d.

In FIG. 10b, the Divider #2, 394, functions in the same manner as previously described for divider 326 of FIG. 10a. Divider 394 receives a # Good Readings' signals via a plurality of conductors 396 from the # Good Readings Counter #2, 266. Divider 394 divides the # Good Readings' value into the SUM' value to generate the final calculated average output temperature signal TXAVG on a plurality of conductors 398. The divider 394 is put into operation by a START DVD #2 signal applied to an enable EN input terminal from a START DVD F/F #2, 400, of FIG. 10d. Flip-flop 400 is enabled to be set by the previously described START DVD signal from the output of AND gate 336 of FIG. 10c and sets when the CLK signal is applied to its trigger T input terminal. Flip-flop 400 is reset by DVD2 COMP binary 1 signal from divider 394 via conductor 402 at the completion of the divide operation.

The DVD2 COMP signal is also applied to a TXAVG REG 404 and to an enable EN input terminal of a second SUBTRACTER 406 of FIG. 10c. As shown in the timing diagram of FIG. 11d, the DVD2 COMP signal clocks the TXAVG signals into register 404 and into the subtracter 406 to enable the subtraction of TXAVG from the temperature base reference TBASE. The TBASE reference value is provided to subtracter 406 from a temperature base reference TBASE REF 408 via a plurality of conductors 410.

Referring to FIG. 10e, the results of the subtraction are gated into a KVAL2 holding register 413 by a DVD2 COMP signal from an inverter 416. The DVD2 COMP signal goes to a binary 1 when the DVD2 COMP signal goes to a binary 0 as shown in FIG. 11d. Register 413 now contains the calculated correction value KVAL2. This value is applied in digital form on a plurality of conductors to a digital to analog (D/A) converter 415 as the KVAL2 signal.

Converter 415 converts the KVAL2 signal to an analog signal KVAL2', the latter being applied to two comparator circuits KVAL2' < 0, 417 and KVAL2' > 0, 419. Each of the comparators 417 and 419 also receives a reference signal from a zero reference source 421 for comparison with the KVAL2' signal. If the KVAL2' signal is less than the reference value (zero), comparator 417 is activated to apply a binary 1 enable signal as one input to an AND gate 423 via a conductor 425. In a similar fashion, if the KVAL2' signal is greater than the reference value, comparator 419 applies a binary 1 signal to one input of a second AND gate 427 via a conductor 429. So long as the value of KVAL2' is equal to the reference value (zero), each of the comparator outputs on conductors 425 and 429 will remain at a binary 0 state preventing either of the AND gates 423 or 427 from being enabled. This would be the condition when there is no difference between the base temperature (TBASE) and the calculated average temperature (TXAVG). So long as any difference exists between these signals (positive or negative) the appropriate one of the comparators 417 or 419 will generate a binary 1 signal.

Reference is now made to a delay one-shot (OS) multivibrator 431 of FIG. 10e receiving the DVD2 COMP signal from inverter 416 of FIG. 10d. The one-shot 431 generates sufficient delay for the signals on conductors 425 and 429, resulting from the D/A conversion of the KVAL2 signals and a comparison of the KVAL2' signal, to settle down before applying the output of one-shot 431 to a second input of each of the AND gates 423 and 427.

When one-shot 431 applies a binary 1 pulse to gates 423 and 427, the gate receiving a binary 1 signal on one of the conductors 425 or 429 will be enabled. If neither one of the comparators 417 and 419 is generating a binary 1 signal, neither of the AND gates will be enabled by the output pulse from one-shot 431.

It will be noted that the outputs of AND gates 423 and 427 are applied to DEC (decrement) and INC (increment) input terminals of an up/down counter (KVALR2 REG) 437 via conductors 433 and 435, respectively. The KVALR2 register 437 is reset to zero at system start-up by the system reset signal on conductor 256, thus KVALR2 will always count up or down from zero. If the value of KVAL2' is greater than the reference (zero), AND gate 427 will be enabled when oneshot 431 is triggered to apply a binary 1 increment signal to the KVALR2 register on conductor 435. In contrast, if the value of KVAL2' is less than the reference, AND gate 423 will be enabled to apply a binary 1 decrement signal to the KVALR2 register.

The increment and decrement signals to the KVALR2 register cause the count in that register to be increased or decreased by a vernier constant as specified by the designer's choice. In the preferred embodiment, the value of KVALR2 is changed by a vernier constant or count of one for each increment or decrement signal applied to the KVALR2 register. Thus, it can be seen that, at the termination of each divide operation by the controller of FIGS. 10a-10e, the value of KVALR2 will be adjusted or not adjusted in accordance with the difference between the values of the TBASE and the TXAVG signals.

The output of the KVALR2 register is applied in digital form on a plurality of conductors 439 to a digital to analog (D/A) converter 441. Converter 441 converts the digital value from the register 437 to an equivalent analog signal KVALR2 on a conductor 443. The KVALR2 signal is applied to a conventional integrator amplifier 445 which provides an integrated output signal KVALR2' to a positive input terminal of a summer 47 via a conductor 449.

The KVALR2 signal from converter 441 is also applied as one input to each of two conventional comparator circuits 451 and 453 via conductor 443 and a conductor 455. Comparators 451 and 453 also each receive at respective input terminals a +LKVAL2 and a -LKVAL2 reference limit signal from corresponding reference sources 457 and 459. The purpose of comparators 451 and 453 is to compare the absolute value of KVALR2 with their respective positive and negative reference limits +LKVAL2 and -LKVAL2.

Whne the value of KVALR2 is within the limits specified by the + and - LKVAL2 signals, the output signal of each of the comparators 451 and 453 is a binary 0. These output signals are applied to an OR gate 461. If the value of KVALR2 exceeds either of the + or - LKVAL2 limits, the corresponding comparator detecting an out of limit condition will apply a binary 1 enable signal to OR gate 461. When OR gate 461 is enabled, it generates a binary 1 signal which may be used to turn on an alarm for the purposes previously described in connection with FIGS. 9a and 9b.

Reference is now made to an analog backup means 36' of FIG. 10e which is generating a mid-value signal ATXDD' in digital form on a plurality of conductors 463. The ATXDD' signals are applied to a conventional subtractor 465 also receiving the TXAVG signals on conductors 467. Subtractor 465 is enabled to subtract the TXAVG value from the ATXDD' value in response to the binary 1 DVD2 COMP signal applied to its enable (EN) input terminal at the time shown in FIG. 11d.

The highly accurate calculated average turbine temperature (TXAVG) is subtracted from ATXDD' (mid-value signal) approximately equal to the average turbine temperature because of the expected differences between these two values. The subtraction yields a second correction value in the form of digital signals on a plurality of conductors 469.

The signals on conductors 469 are gated into an output register 471 by the DVD2 COMP binary 1 signal from inverter 416.

The output of register 471 is provided via a plurality of conductors 418 to a conventional digital-to-analog (D/A) converter 420 as a correction value KVAL1. Converter 420 converts the KVAL1 signals to an analog signal KVAL1' for application to a positive terminal of a summer 473.

The output of summer 473 is applied to a conventional integrator amplifier 422 which generates an integrated fuel correction signal KVALR1 on a conductor 475 to one input terminal of summer 447. The KVALR1 and KVALR2' signals are algebraically combined in summer 447 with the aforementioned DELR1 signal on a conductor 477 to generate the temperature trim signal BTMTEM on conductor 52.

The DELR1 delta reference signal is generated by a summer 479 receiving the ATREF signal (safe temperature reference) and a TBASE' signal having a value proportional to the turbine base temperature or set point. As previously described, the DELR1 signal has a value proportional to the difference between the ATREF and TABSE' signals and serves as a temperature bias reference for maintaining the turbine temperature at a desired value.

The TBASE' signal is analog in nature and is generated by a digital-to-analog (D/A) converter circuit 481 which converts the digital signals TBASE on conductors 410 supplied by the TBASE REF source 408.

The BTMTEM signal is combined in the analog means 36' with the ATXDD and ATREF signals to generate the temperature fuel control signal TFC on conductor 58 for application to the low value gate 34 of FIG. 1.

The KVALR1 signal from amplifier 422 is fed back to a negative input terminal of summer 473 where that signal is combined with the KVAL1' signal for the same reason previously described in connection with amplifier 257 of FIG. 9b.

The # Good Readings' signals from the output of counter 266 of FIG. 10b are applied via conductors 424 to a # Good Readings< 8 decode logic 426 of FIG. 10d. Decode 426 functions in the same manner as previously described for the Decode Logic 236 of FIG. 9b. Whenever the number of good readings in counter 266 is less than 8, the decode 426 generates a binary 1 output signal to an AND gate 428. AND gate 428 also receives the DVD2 COMP signal from inverter 416. As previously described in the previous embodiments, if the number of good readings is less than 8, it is deemed necessary to place the turbine under control of the analog backup means. When the DVD2 COMP signal goes to a binary 1 and if the output of decode 426 is a binary 1, AND gate 428 is enabled to apply a binary 1 RESET INT signal to integrators 422 and 445 causing the BTMTEM signal to be ramped to zero, thus causing the TFC signal to take on the value of the algebraic sum of the ATREF and ATXDD signals.

Reference is now made to FIG. 12, which is a modified form of the temperature analog backup means 36' for use in the embodiment of FIGS. 10a-10e. The temperature analog backup means of FIG. 12 contains like numerals with prime designations corresponding to those components previously described in connection with FIGS. 1 and 2. For example, the mid-value selector 64 of FIG. 2 is shown as 64' in FIG. 12. The analog backup means 36' of FIG. 12 is basically the same as that previously described in connection with FIG. 2, with the exception of the inclusion of an A/D converter 432 and an ATXDD holding register 434. In the embodiment of FIG. 12, the selector 64' continuously monitors the AOTA through AOTC turbine sensor signals as previously described to generate an analog mid-value signal shown as ATXDA for input to the A/D converter 432. The A/D converter 432 functions similarly to that described in connection with the A/D converter 274 of FIG. 10a. Converter 432 receives the CLK clock signal from generator 250 and the START DVD signal from the output of AND gate 336 of FIG. 10c to reset and start that converter when the START DVD signal goes to a binary 1. At the termination of an A/D conversion, the output of the converter 432 is a digital value representative of the mid-value signal shown on a plurality of conductors 436 as the ATXDD signals. The ATXDD signals are applied to the input of the ATXDD holding register 434. At the completion of the divide by the divider 394, the DVD2 COMP signal on conductor 402 goes to a binary 1 gating the ATXDD signals into register 434. Also at this time, the contents of the register 434 are gated as signals ATXDD' on conductors 463 into the subtracter 465 by the DVD2 COMP signal (see FIG. 10e).

Still referring to FIG. 12, the ATXDA signal, in conjunction with the ATREF' and BTMTEM signals, is applied to summer 66' to provide the temperature fuel control signal TFC through amplifier 68' as previously described in connection with FIG. 2. In normal operation, the output signal TFC on conductor 58 from amplifier 68' is representative of the algebraic sum of the ATXDA, ATREF' and BTMTEM signals. However, when the number of good readings is less than 8, as decoded by decode 426, AND gate 428 is enabled to generate the RESET INT signal to reset the integrators 422 and 445 causing the output signal BTMTEM to ramp down to zero. When this occurs, the analog backup means 36' takes over in the manner as previously described to now provide the TFC signal to the turbine through the low value gate 34. In this latter instance, the value of the TFC signal is representative of the algebraic sum of the ATXDA and ATREF' signals since the BTMTEM signal now have an ineffective value.

The comparator and latch circuits for generating the fuel trip signal on conductor 60 (see FIG. 2) are not shown in FIG. 12. These circuits have been left off of FIG. 12 to simplify the drawing; however, it is to be understood that those circuits could be connected to the output of the selector 64' in the same manner as shown in FIG. 2.

While illustrative embodiments of the invention have been described herein, modifications and adaptations thereof may be made by those skilled in the art without departure from the spirit and true scope of the invention as defined by the appended claims.

Patent Citations
Cited PatentFiling datePublication dateApplicantTitle
US3709626 *Sep 16, 1971Jan 9, 1973Gen ElectricDigital analog electrohydraulic turbine control system
US3719071 *Aug 30, 1971Mar 6, 1973Avco CorpMeasurement of gas temperature variations in a gas turbine engine
US3764785 *Oct 29, 1970Oct 9, 1973United Aircraft CorpTurbine power plant control system
US3789665 *Feb 22, 1972Feb 5, 1974Avco CorpInferred measurement of the turbine inlet temperature of a gas turbine engine
US3899878 *Feb 23, 1973Aug 19, 1975Int Harvester CoApparatus for indicating gas temperatures
US3955359 *Jun 20, 1973May 11, 1976Westinghouse Electric CorporationBearing temperature system failure detection apparatus suitable for use in power plants and like apparatus
US4058975 *Dec 8, 1975Nov 22, 1977General Electric CompanyGas turbine temperature sensor validation apparatus and method
Referenced by
Citing PatentFiling datePublication dateApplicantTitle
US4525783 *Dec 5, 1983Jun 25, 1985Robert Bosch GmbhMethod and apparatus for determining the individual manipulated variables of an internal combustion engine, and in particular of a gas turbine
US6633828Mar 21, 2001Oct 14, 2003Honeywell International Inc.Speed signal variance detection fault system and method
US6804600Sep 5, 2003Oct 12, 2004Honeywell International, Inc.Sensor error detection and compensation system and method
US6885923Jul 24, 2003Apr 26, 2005Honeywell International Inc.Speed signal variance detection fault system and method
US6960900 *Nov 28, 2003Nov 1, 2005General Electric CompanyMethod and apparatus for starting a gas turbine using a polyphase electric power generator
US9002617 *Jul 10, 2013Apr 7, 2015General Electric CompanyGas turbine engine controller with event trigger
US9207128Jun 5, 2012Dec 8, 2015General Electric CompanyDynamic fiber temperature sensing package and method of assembling the same
US20050116689 *Nov 28, 2003Jun 2, 2005General Electric CompanyMethod and apparatus for starting a gas turbine using a polyphase electric power generator
US20150019104 *Jul 10, 2013Jan 15, 2015General Electric CompanyGas turbine engine controller with event trigger
Classifications
U.S. Classification60/773, 60/39.281, 701/100
International ClassificationF02C9/32
Cooperative ClassificationF05D2260/80, F05D2270/303, F02C9/28
European ClassificationF02C9/28